http://pubs.acs.org/cgi-bin/sample.cgi/ancham/2007/79/i12/html/ac070719q.html

Water Analysis: Emerging Contaminants and Current Issues
August, 2007
Susan D. Richardson

Drinking Water Disinfection Byproducts

In addition to new regulations and rules involving DBPs (e.g., the Stage 2 Disinfectants/DBP Rule and the UCMR-2, which requires monitoring of nitrosamines), there are also new, emerging issues with DBPs (91). New human exposure research is revealing that inhalation and dermal exposures (from showering, bathing, swimming, and other activities) can provide equivalent exposures or increased exposures to certain DBPs (91). Therefore, these exposure routes are now being recognized in new epidemiologic studies that are being conducted. And, epidemiology studies are beginning to focus more on reproductive and developmental effects-which recent studies have been shown to be important. A recent review article outlines these important routes of exposure, along with new, emerging DBPs (91).

Toxicologically Important DBPs. Toxicologically important DBPs include brominated, iodinated, and nitrogen-containing DBPs (so-called "N-DBPs"). Brominated DBPs are more carcinogenic than their chlorinated analogues (91), and new research is indicating that iodinated compounds may be more toxic than their brominated analogues (91). Brominated and iodinated DBPs form due to the reaction of the disinfectant (such as chlorine) with natural bromide or iodide present in source waters. Coastal cities, whose groundwaters and surface waters can be impacted by salt water intrusion, and some inland locations, whose surface waters can be impacted by natural salt deposits from ancient seas or oil-field brines, are examples of locations that can have high bromide and iodide levels. A significant proportion of the U.S. population and several other countries now live in coastal regions that are impacted by bromide and iodide; therefore, exposures to brominated and iodinated DBPs are important. Early evidence in epidemiologic studies also gives indication that brominated DBPs may be associated with the new reproductive and developmental effects, as well as cancer effects.

Specific DBPs that are of current interest include iodo acids, bromonitromethanes, iodo-THMs, brominated forms of MX (MX is 3-chloro-4-(dichloromethyl)-5-hydroxy-2(5H)-furanone), haloaldehydes, haloamides, and NDMA, which is not brominated, but is classified as a probable carcinogen (91). Iodoacetic acid, one of five iodo acids identified for the first time in chloraminated drinking water, has recently been shown to be more genotoxic and cytotoxic to mammalian cells than all DBPs that have been studied, including the regulated HAAs and bromate (91). It is a factor of 2? more genotoxic than bromoacetic acid, which is the most genotoxic of the regulated HAAs. Other iodo acids identified-bromoiodoacetic acid, (Z)-3-bromo-3-iodopropenoic acid, (E)-3-bromo-3-iodopropenoic acid, and (E)-2-iodo-3-methylbutenedioic acid (91)-have been synthesized and are currently under investigation for genotoxic and cytotoxic effects. They were initially discovered in chloraminated drinking water extracts using methylation with GC/high-resolution-EI-MS. Analytical methods for the five iodo acids have been developed for a current occurrence study to determine their concentrations in chloraminated drinking water. These iodo acids are of concern not only for their potential health risks but also because early research indicates that they may be maximized (along with iodo-THMs) in waters treated with chloramines. Chloramination has become a popular alternative to chlorination for plants that have difficulty meeting the regulations with chlorine, and its use is expected to increase with the advent of the new Stage 2 D/DBP Rule. Chloramines are generated from the reaction of chlorine with ammonia, and it appears that the length of free chlorine contact time (before ammonia addition to form chloramines) is an important factor in the formation of iodo acids and iodo-THMs. Because of chlorine's competing reaction to form iodate as a sink for the natural iodide, it is likely that plants with significant free chlorine contact time before the addition of ammonia will not produce substantial levels of iodo acids or iodo-THMs.

The bromonitromethanes (including dibromonitromethane, tribromonitromethane, and bromonitromethane) are extremely cytotoxic and genotoxic to mammalian cells (91). Dibromonitromethane is at least 1 order of magnitude more genotoxic to mammalian cells than MX and is more genotoxic than all of the regulated DBPs, except for monobromoacetic acid. Bromonitromethanes have been found to be DBPs formed by chlorination or chloramination and have been shown to increase in formation when preozonation is used before chlorine or chloramine treatment. Bromonitromethanes, iodo-THMs, and brominated forms of MX (so-called BMXs), as well as other priority DBPs were the focus of a U.S. Nationwide DBP Occurrence Study, which was recently published in Environmental Science & Technology (92). This Nationwide Occurrence Study focused on approximately 50 priority DBPs that were selected from an extensive prioritization effort (according to predicted cancer effects). In this study, haloacetaldehydes represented the third major class formed on a weight basis (behind THMs and HAAs). An important finding was that while the alternative disinfectants significantly lowered the formation of regulated DBPs (THMs and HAAs), other high-priority DBPs were increased in formation with alternative disinfectants. For example, iodinated DBPs (iodo-THMs and iodo acids) were increased in formation with chloramination, dichloroacetaldehyde was highest at a plant using chloramines and ozone, and preozonation was found to increase the formation of halonitromethanes. This has important implications for drinking water treatment, as many plants in the United States have switched or are switching to alternative disinfectants to meet the Stage 1 and 2 D/DBP Rule requirements. This study also reports the highest levels of MX analogues to-date, with MX analogues and brominated MX analogues frequently being found at levels above 100 ng/L, and in two plants the sum of these analogues was at low-ppb levels. Finally, 28 new, previously unidentified DBPs were reported, including brominated and iodinated acids, a brominated ketone, and chlorinated and iodinated aldehydes. Despite the fact that more than 90 DBPs were measured in this study, only about 30 and 39% of the total organic halide (TOX) and total organic bromine (TOBr) were accounted for, respectively, by the sum of the measured DBPs. This is consistent with earlier studies that have shown that there is more TOX accounted for in chlorinated drinking water, as compared to drinking water treated with alternative disinfectants.

GC/MS continues to be an important tool for measuring DBPs and identifying new DBPs. However, LC/MS is being increasingly used for highly polar DBPs and high molecular weight DBPs. In fact, this was the focus of a recent review article by Zwiener and Richardson (93). Useful derivatization techniques, as well as related MS techniques, such as ESI-high field asymmetric waveform ion mobility spectrometry (FAIMS)-MS, IC-ESI-MS, and membrane introduction MS (MIMS), are also discussed. This review covered not only traditional DBPs that are formed by the reaction of the disinfectant (oxidant) with NOM but also newly identified DBPs that are formed by the reaction of the disinfectant with contaminants. Examples of those include reaction products with estrogens, alkylphenol ethoxylates, pesticides, and algal toxins.

Brominated and iodinated DBPs have been the focus of several new studies. In an innovative study, Becalski et al. investigated the potential formation of iodoacetic acids during cooking (94). In this study, municipal chlorinated tap water (containing NOM) was allowed to react with iodized table salt (containing potassium iodide) and with potassium iodide itself in boiling water. Samples were extracted with TAME and methylated prior to analysis with GC/MS. Iodoacetic acid and chloroiodoacetic acid were identified as byproducts, and iodoacetic acid was formed at 1.5 g/L levels when the water was boiled with 2 g/L iodized table salt. The concentration of chloroiodoacetic acid was estimated to be 3-5? lower. Hua et al. examined the effect of bromide and iodide on the formation of DBPs during chlorination (95). TOBr, total organic iodine (TOI), and total organic chlorine (TOCl) were measured in this study, as well as THMs, HAAs, and TOX. At higher levels of bromide, there was a decreasing level of unknown TOX and unknown TOCl, but an increasing level of unknown TOBr. The extent of iodine substitution was much lower than bromine substitution because a substantial amount of iodide was oxidized by chlorine to iodate. The tendency toward iodate formation resulted in the unusual situation where higher chlorine doses actually reduced levels of iodinated DBPs. However, this is not the case with chloramination, where iodo-DBPs preferentially form instead of iodate (92). The method for TOCl, TOBr, and TOI analysis is described in a separate paper by Hua and Reckhow (96). After investigating different pyrolysis-IC procedures, the optimum method included a pyrolytic analyzer that uses pure O2 and offline IC combined with a standard TOX carbon (coconut-based). This procedure allowed the most complete recovery of TOCl, TOBr, and TOI. Brominated and chlorinated acetaldehydes were the focus of another study by Koudjonou and Lebel (97). These DBPs were measured in Canadian drinking water with GC/electron capture detection (ECD), and their stability was investigated. Most of the haloacetaldehydes were found in the drinking waters, with chloral hydrate (trichloroacetaldehyde) comprising 7-51% of the total haloacetaldehydes measured, as well as a substantial portion of the total DBPs (as in the U.S. Nationwide Study). Mixed results were obtained for their stability in drinking water-the trihaloacetaldehydes degraded somewhat over time to the corresponding THMs at increasing pH and temperature.

New DBPs continue to be identified. Often, low- and high-resolution EI-MS is used, and sometimes combinations of GC/MS or LC/MS with Fourier transform (FT)-infrared (IR) spectroscopy or NMR are used. In addition, derivatizing agents continue to be developed to aid in the identification of highly polar DBPs, which are largely unaccounted for. Gong et al. used FT-IR spectroscopy, EI-MS, 1H and 13C NMR spectroscopy, and single-crystal X-ray diffraction to identify a new DBP in chlorinated drinking water (98). This DBP was identified as 2,2,4-trichloro-5-methoxycyclopent-4-ene-1,3-dione. Ames test results showed it to be highly mutagenic. Vincenti et al. tested four newly developed fluorinated chloroformate derivatizing agents for identifying highly polar alcohol, carboxylic acid, and amine DBPs in drinking water with GC/negative chemical ionization (NCI)-MS (also referred to as electron capture negative ionization) (99). 2,2,3,3,4,4,5,5-Octafluoro-1-pentyl chloroformate performed the best, with good reaction efficiency, good chromatographic and spectroscopic properties, low detection limits (10-100 fmol), and a linear response over more than 2 orders of magnitude. The entire procedure from raw aqueous sample to ready-to-inject hexane solutions of the derivatives required less than 10 min. This method was used to identify three highly polar ozonation byproducts: malic acid, tricarballylic acid, and 1,2,3-benzenetricarboxylic acid.

Other Occurrence Studies. Huang et al. used GC/high-resolution-EI-MS to comprehensively identify DBPs formed by the ozonation of polluted source waters (100). Fifty-nine different organic compounds were identified, including low molecular weight carboxylic acids, benzoic compounds, aldehydes, bromoform, bromoacetic acid, dibromoacetic acid, 2,4-dibromophenol, and dibromoacetonitrile. When the NOM was fractionated from the source water into humic acid and hydrophilic neutral fractions, different distributions of DBPs were observed in the fractions. Malliarou et al. recently carried out a large survey of HAAs in UK drinking waters (101). Means ranged from 35 to 95 g/L, and a maximum of 244 g/L was observed. In two out of the three regions investigated, there was a high correlation between total THMs and total HAAs, and the ratio of total THMs to total HAAs was significantly correlated with temperature, pH, and free and total chlorine. This study is particularly important because HAAs are rarely measured in Europe, and most epidemiologic studies relate effects back to THMs only. Another large survey was carried out in Athens, Greece, over a 2-year period (102). DBPs measured (by GC/MS) included THMs, haloacetonitriles, haloketones, chloral hydrate, chloropicrin, and nine HAAs. All DBPs were identified in prechlorinated drinking water samples. The most commonly detected DBPs were chloroform, trichloroacetic acid, dichloroacetic acid, and chloroacetic acid. Annual mean concentrations ranged from 1.1 to 61.8 g/L.

Discovery Research for High Molecular Weight DBPs. More than 50% of the TOX formed in chlorinated drinking water remains unidentified, and much higher percentages of TOX are unaccounted for when alternative disinfectants are used (ozone, chloramine, chlorine dioxide). Earlier ultrafiltration studies indicate that >50% of the TOX in chlorinated drinking water is >500 in molecular weight, which would be missed with traditional GC/MS approaches. ESI-MS/MS is allowing researchers to investigate these high molecular weight DBPs. Most of this work is very preliminary, due to the complexity of the mass spectra obtained ("a peak at every mass" situation). Minear's group at the University of Illinois has carried out much of the pioneering work in this area. In a follow-up study to their earlier work, Zhang and Minear used radiolabeled chlorine (36Cl) to further probe high molecular weight DBPs formed upon chlorination of drinking water (103). Results of this study showed that oxidation was the dominant reaction compared to halogenation and that high molecular weight DBPs decreased when the chlorine contact time was increased. High molecular weight DBPs could not be separated into discrete LC peaks.

NDMA and Nitrosamines. NDMA is a probable human carcinogen, and NDMA and other nitrosamines were recently discovered to be DBPs in drinking water. NDMA can form in chloraminated or chlorinated water. 15N-Labeling studies have shown that the nitrogen present in monochloramine becomes incorporated into the structure of NDMA. And, as with iodo-DBP formation, the length of free chlorine contact time prior to ammonia addition to form chloramines can be an important factor in the formation of NDMA. Charrois and Hrudey published a recent study showing that a free chlorine contact time of 2 h (before ammonia addition) resulted in significant reductions (up to 93%) in NDMA formation (104). Chlorination can also form NDMA, when nitrogen precursors are present (e.g., natural ammonia in the source water or nitrogen-containing coagulants, such as diallyldimethylammonium chloride, used in water treatment). NDMA was initially discovered in chlorinated drinking waters from Ontario, Canada, and has since been found in other locations. The detection of NDMA in U.S. waters is largely due to improved analytical techniques that have allowed its determination at low-nanogram per liter concentrations. NDMA is generally present at low nanograms per liter in chloraminated/chlorinated drinking water, but it can be formed at much higher levels in wastewater treated with chlorine. Following its discovery in California well water, the State of California issued an action level of 0.002 g/L (2 parts per trillion) for NDMA, which was subsequently revised to 0.01 g/L, due to the analytical difficulty in measuring it at the original proposed level (www.dhs.ca.gov/ps/ddwem/chemicals/NDMA). NDMA is not currently regulated in the United States for drinking water, but is now included on the UCMR-2, where occurrence data are being collected on a national scale for NDMA and other nitrosamines. Ontario has issued an interim maximum acceptable concentration for NDMA at 9 ng/L (www.ene.gov.on.ca/envision/gp/4449e.pdf). Andrzejewski et al. published a nice review on NDMA in 2005, where its toxicological issues, mechanisms of formation in drinking water treatment, and physiochemical properties are discussed (105). This review also cites the possibility of NDMA being formed with chlorine dioxide disinfection.

To-date, all methods to measure NDMA have been GC/MS(/MS) or GC/ECD methods, including the EPA method created to measure nitrosamines (EPA Method 521). Zhao et al. created the first LC/MS/MS method to measure nitrosamines and, in doing so, identified two new nitrosamine DBPs in drinking water-nitrosopiperidine and nitrosodiphenylamine (106). LC/MS/MS was essential for detecting nitrosodiphenylamine, as it is thermally unstable and cannot be measured by GC/MS. An isotopically labeled NDMA standard was used as the surrogate standard for determining recovery, and isotopically labeled N-nitrosodi-n-propylamine was used as an internal standard for quantification. Detection limits ranged from 0.1 to 10.6 ng/L. Measurements in a drinking water distribution system revealed that nitrosamine concentrations increased with increasing distance from the water treatment plant, indicating that the amount of formation was greater than the amount of decomposition. Cheng et al. expanded and refined three previously existing GC/MS/MS methods for measuring nitrosamines in drinking water, wastewater, and recycled water (107). Detection limits for two SPE-GC/MS/MS methods ranged from 0.3 to 1.4 ng/L, and detection limits for a micro-liquid-liquid extraction-GC/MS/MS method ranged from 2 to 4 ng/L. These methods were used to measure NDMA and several other nitrosamines in drinking water, wastewater, and recycled water in California. In drinking water, NDMA was the only nitrosamine detected, but other nitrosamines were present in recycled water and wastewater. Cha et al. reported a new LC-fluorescence method for measuring NDMA in water (108). Samples were denitrosated and derivatized with dansyl chloride for fluorescence detection. Detection limits of 10 ng/L could be achieved. This method did not suffer interferences even in complex wastewater samples. Grebel et al. developed a new SPME method for extracting seven nitrosamines from water (109). SPME could be used with nitrogen chemiluminescence detection, nitrogen-phosphorus detection, or chemical ionization (CI)-MS. Detection limits for NDMA ranged from 30 to 890 ng/L.

Mechanistic Studies. Researchers continue to explore the mechanism of formation of nitrosamines. Schreiber and Mitch examined the importance of chloramine speciation and dissolved oxygen on the formation of nitrosamines (110). Dichloramine and dissolved oxygen were found to be critical in their formation, and a new nitrosamine formation pathway was proposed, in which dichloramine reacts with secondary amine precursors to form chlorinated dialkylhydrazine intermediates. Oxidation of these intermediates by dissolved oxygen to form nitrosamines competes with their oxidation by chloramines. This new model was able to explain the formation of nearly all nitrosamine species. Chen and Valentine developed a kinetic model to validate proposed reactions and predict NDMA formation in chloraminated drinking water (111). Inputs to this model include chloramine demand, a coefficient relating the amount of NDMA produced to the amount of NOM oxidized, and other kinetic parameters describing NOM oxidation. NOM oxidation was determined to be the rate-limiting step governing NDMA formation.

Mechanistic studies have also been carried out for other DBPs, including cyanogen chloride, N-chloroaldimines, and ozonation DBPs. Lee et al. examined 17 amino acids as potential precursors for CNCl in chlorinated drinking water (112). Among these amino acids, only glycine was found to produce detectable CNCl, and the glycine nitrogen was stoichiometrically converted to CNCl at pH <6. From examinations of river water, it was estimated that glycine may account for 42-45% of the CNCl formed (at pH 8.2). In another study by Freuze et al., amino acids were investigated as precursors to DBPs involved in an odor episode in Paris (113). The reaction of several amino acids with chlorine was investigated to solve the odor mystery. N-Chloroaldimines were identified in these amino acid-chlorine reactions by GC/MS, and they were suspected of being the DBPs responsible for the odor episode. Finally, These and Reemtsma used size exclusion chromatography with Q-TOF-MS to examine ozone DBP formation of different NOM fractions (114). A preferential reaction with fulvic acids at a low oxidation state (low O/C ratio) and a high degree of unsaturation (low H/C ratio) was observed, and the data suggested that molecules with a more extended carbon skeleton and fewer carboxylate substituents are more reactive with ozone.

Other New DBP Methods. Several new methods have been developed for the measurement of DBPs (beyond nitrosamines mentioned earlier). Khan et al. reported a new aqueous-phase aminolysis method to measure epoxides in water (115). This method also uses SPE, silylation, and GC/MS analysis. With this method, 1,2-epoxybutane, epichlorohydrin, and epifluorohydrin could be measured at 5-10 ng/L detection limits. Onstad and Weinberg created a refined method using liquid-liquid extraction, methylation, and GC with micro-ECD or ion trap-MS detection for measuring halogenated furanones (MX analogues) in drinking water (116). A preconcentration factor of 1000:1 allowed low-nanogram per liter detection limits. This method was used to measure the 12 halogenated furanone species in the U.S. Nationwide Study discussed earlier. Yang and Shang created a new MIMS method to quantify CNCl and cyanogen bromide in water (117). A linear response over 3 orders of magnitude was achieved, and CNCl and CNBr could be measured down to limits of 1.2 and 3.8 g/L, respectively. Recoveries were >93%. A new SPME-GC/ECD method to measure 2,4,6-trichloroanisole in chlorinated drinking water was also developed (118). Detection limits and quantification limits of 0.7 and 2.5 ng/L, respectively, were achieved. THMs could also be measured with this method. De Borba et al. created a new IC method to measure bromate in municipal and bottled drinking waters (119). This method utilized an electrolytically generated hydroxide eluent combined with a hydroxide-selective anion-exchange column and was able to provide significant noise reduction, along with 0.5 g/L detection limits.

Several new continuous, online methods have also been recently developed, and these have the promise of being used in water treatment plants to allow real-time determination of DBPs. Wang et al. developed a new continuous hollow-fiber, liquid-liquid membrane extraction-LC/UV method to measure HAAs in drinking water (120). Method detection limits were at sub-ppb levels. Simone et al. developed an online IC method for HAAs that uses a postcolumn reaction with nicotinamide and fluorescence detection (121). Detection limits of 0.5-5 g/L were achieved, and this on-line method was compared to EPA Method 552.3. Finally, Brown and Emmert developed a new on-line method for THMs, using capillary membrane sampling and GC/ECD detection (122). Method detection limits were in the 0.5 g/L range. This method was compared to EPA Method 502.2, and it offers advantages for monitoring a drinking water distribution system because it is a near real-time method and can be used at remote locations in the distribution system.

New Human Exposure Studies. Researchers have been investigating other routes of exposure, besides ingestion, in new human exposure studies of drinking water DBPs. And, in many cases, inhalation and dermal exposures that would result from bathing or showering offer greater exposures to particular DBPs than ingesting 2 L of water per day. Exhaled breath is often a convenient, noninvasive way to assess a person's exposure, either dermally or through inhalation. Once a DBP has been absorbed either through the lungs or through the skin, it is transported to the blood stream, where it can be released in exhaled breath from the lungs. Blood measurements are more invasive, but can be more precise measures of exposure. It is of particular interest to epidemiologic studies to know the entire dose of specific DBPs being investigated for effects. Xu and Weisel investigated the dermal absorption of 1,1-dichloropropanone, 1,1,1-trichloropropanone, and chloroform in human volunteers (in their exhaled breath) following a 30-min bath (123). The maximum haloketone breath concentration ranged from 0.1 to 0.9 ug/m3, which were approximately 2 orders of magnitude lower than the maximum chloroform breath concentrations. The permeability of chloroform was found to be much higher than the permeability of the haloketones. Gordon et al. carried out a human exposure study that investigated breath and blood THM levels from 12 common household water-use activities (124). Water, indoor air, blood, and exhaled breath samples were collected during each exposure activity. Although showering (10 min), bathing (14 min), machine washing of clothes, and opening dishwashers at the end of the cycle resulted in significant increases in indoor air chloroform levels, only showering and bathing caused significant increases in breath chloroform levels. For bromodichloromethane, only bathing produced significantly higher concentrations. For chloroform from showering, strong correlations were observed for indoor air and exhaled breath, blood and exhaled breath, indoor air and blood, and tap water and blood. Evidence of the importance of dermal and inhalation routes for DBPs, a new epidemiologic study by Villanueva et al. revealed a higher risk of bladder cancer from showering, bathing, and swimming in pools (125). Long-term THM exposure was associated with a 2-fold bladder cancer risk (odds ratio of 2.10) for average household THM levels of >49 g/L. The odds ratio for ingestion was 1.35 (compared to people who did not drink tap water), and the odds ratio from showering and bathing was 1.83.

New Swimming Pool Research. Related to other research involving alternate exposures to ingesting drinking water, swimming pool studies have shown a marked increase in the last 2 years. The Villanueva et al. epidemiologic study mentioned earlier showed an odds ratio of 1.57 for swimming in pools and developing bladder cancer (125). Zwiener et al. published a review article on swimming pool waters, detailing the adverse health effects (including asthma, bladder cancer, and endocrine effects), the formation of DBPs in swimming pool water, and precursor chemicals that give rise to them (126). Details on swimming pool operation and treatment are also given. Glauner et al. investigated the elimination of swimming pool DBPs using ozonation and advanced oxidation processes (ozone/UV and ozone/hydrogen peroxide) (127). Advanced oxidation processes (AOPs) substantially reduced the levels of TOC, adsorbable organic halogen, and THMs. A contact time of 3 min between the pool water and the oxidants was found to be sufficient for lowering DBP levels. Ozonation showed a small advantage to AOPs in removing THMs, and the combination of membrane filtration and AOPs resulted in the elimination of 10-90% of the DBPs and their precursors. The ozone/hydrogen peroxide process was recommended for pool water treatment because of higher elimination rates compared to ozonation alone and lower costs as compared to ozone/UV treatment.